1. Field of the Invention
The present invention relates, in general, to an improved optical waveguide and method of making same and, in particular, to an improved signal processing device such as an Integrated Optical Spectrum Analyzer.
2. Description of the Prior Art
Optical waveguides formed by the thermal in-diffusion of titanium into the surface of Lithium Niobate (LiNbO.sub.3) crystals have become the basis for numerous advanced optical guided wave signal processing devices such as the integrated optics spectrum analyzer (IOSA).
Titanium in-diffused waveguides are capable of propagating optical waves with little attenuation in a region very close (within 3 .mu.m) to the crystal surface where the optical beam can interact efficiently with high frequency surface acoustic waves (SAW) through the acousto-optic effect or with electric fields generated by metal surface electrodes via the electro-optic effect. However, several problems attend the close proximity of the guided wave to the surface of the crystal.
Surface roughness, micro-scratches and contamination can increase the amount of in-plane scattering and thereby reduce the dynamic range of the signal processor. Coupling light efficiently into and out of the waveguide is also more difficult to achieve when the waveguide is tightly confined to the crystal surface. Laser diodes, which are often the light sources for optical guided wave signal processors, have highly divergent beams which necessitate extremely accurate and close positioning between the active region of the diode and the well polished edge of the waveguide. Conversely, guided beams exiting the waveguide diverge quickly and, therefore, precise butt coupling of a detector to the waveguide edge is essential in order to capture the maximum amount of light.
Titanium in-diffused waveguides in LiNbO.sub.3 can propagate transverse electric (TE) and transverse magnetic (TM) polarized modes simultaneously. The simultaneous presence of two types of modes leads to the deterioration of the performance of geodesic lenses since the lenses will exhibit a different focal length for each mode type. Furthermore, geodesic lenses, needed either for collimating or focussing, operate with light rays at large angles (65.degree.-90.degree.) with respect to the optical axis of LiNbO.sub.3 . In this region, both TE and TM modes are leaky modes with TE modes being dominant. It is desirable to eliminate TM modes in order to suppress the occurrence of double focussing in the detector region and reduce the noise floor, and, thus, improve the dynamic range of the device.
Finally, these waveguides are susceptible to optical damage at high optical power densities especially for visible light wavelengths. This usually occurs at the input coupling region where the light beam often has its smallest cross-section. Consequently, the maximum amount of optical power than can be coupled into the guide is restricted to the damage threshold.
The problem of in-plane scattering caused by surface irregularities in the planar in-diffused waveguides has not yet been solved in a clear and consistent manner. However, careful control of several factors in the waveguide preparation has been shown to reduce the level of scattering. These factors include: the thickness of the pre-diffusion titanium layer on the crystal, the duration and temperature of the diffusion and the influx of argon and oxygen gases during the diffusion. Careful surface polishing may also aid in lowering scattering but this is not always successful or consistent.
A variety of techniques have been developed for coupling light into the in-diffused waveguides. Some of the more efficient techniques include: the direct butt-coupling of a laser diode to a polished waveguide edge and the use of lenses to focus the light onto the polished edge. Both methods require very precise positioning of the optical elements relative to the waveguide edge.
Elimination of TM polarized waveguide modes can be accomplished with a tapered transitional waveguide situated between a high refractive index and a low refractive index waveguide while ensuring the guided beam is incident at the transition junction at the Brewster angle. Methods for removing TM modes propagated in the geodesic lenses have not as yet been published.
The problem of optical damage in LiNbO.sub.3 in-diffused waveguides has not been solved for visible light wavelengths, although doping the crystal with MgO during manufacture or adding water vapour to the argon gas during in-diffusion have been reported to help.
The previous proposed solutions to the planar scattering problem are neither definitive nor necessarily consistent when implemented because of the many variables involved in waveguide fabrication processes, while solutions to the non-planar scattering have yet to be put forward.
The coupling problem has not been solved in an optimal sense. Although efficient edge coupling of light to the waveguide can be attained, it is at the cost of greater complexity and precision since the available coupling region is only 3 .mu.m thick. Direct coupling of a laser diode or photodetector results in a compact device but it is a difficult task in practice. On the other hand, while the use of focussing optics makes the positioning requirements less stringent, they increase the size and bulk of the signal processor.
Removal of filtering or TM modes by means of transitional waveguides is complicated and impractical in practice. Furthermore, imperfect waveguide matching may cause unacceptably high levels of scattering loss. Solutions to the problem with TM modes in geodesic lenses have not yet been advanced.
Techniques for increasing, in the visible light range, the optical damage resistance in titanium in-diffused waveguides have been successful but the resistance is still lower than that of out-diffused waveguides in LiNbO.sub.3 .
In general, if all the previously described solutions were incorporated into a single guided wave signal processing device, the complexity and cost of such a device would likely increase and the production yield fall and, in any case, there is no assurance that such solutions would function in concert and improve the performance of the device. Therefore, it can be stated that, at present, there is no single technique or system which will alleviate all the aforementioned problems in a simple and reproducible fashion.